For half of people, function holds until the last two years

Design

Harmonised multi-cohort longitudinal study with group-based trajectory modelling, 1997-2016

Population

2,294 Swedish decedents who died aged 70 or over, 62% women, mean age at death 89.8 years

Primary outcome

Activities of daily living limitations across four intervals before death

Effect

Three trajectories - constantly low 53%, low then sharply increasing 36%, high-increasing 11%; women, lower education and older age at death associated with the worse trajectories

Six Swedish longitudinal ageing studies running from 1997 to 2016 were harmonised to follow 2,294 people who died aged 70 or over and had activities of daily living measured on at least two occasions. Mean age at death was 89.8 years and 62% were women. Limitations were scored 0 to 5 across four windows before death: 0-2 years, more than 2 to 4, more than 4 to 6, and 6 to 12.

Group-based trajectory modelling found three patterns. Just over half - 53% - stayed at constantly low disability throughout. Another 36% were low and then rose sharply, and 11% were already high and got worse. Women, people with less education and people who died at older ages were more likely to be in the rising or persistently high groups, and the high-increasing trajectory was associated with cognitive impairment.

The headline number is the useful one for a conversation with families: for most people, everyday function is preserved until roughly the final two years of life, and the decline when it comes is compressed rather than gradual. That contradicts the widely held expectation of a long slow slide, and it has planning consequences in both directions - it argues against assuming a decade of dependency, and it argues for recognising that when the slope does change, the window to arrange care is short.

• Use the trajectory framing in family conversations: for most, function holds and then falls quickly
• A sharp change in activities of daily living is a signal to start care planning, not to wait for confirmation
• Cognitive impairment marked the group that was already disabled and deteriorating further
• Lower education and female sex were associated with the worse trajectories in this Swedish cohort
• Mean age at death was 89.8 - these are the oldest old, and the pattern may differ at younger ages

Why it matters

The common expectation of a long gradual decline is wrong for most people, and that changes when care planning should start.

Don't overread it

Trajectory groups are a modelling choice, not discovered categories, and the cohort was selected on having died at a mean age of 89.8.

The statistics, in plain English

Group-based trajectory modelling does not discover real groups; it fits a chosen number of curves to continuous data and assigns people to whichever fits best. Three trajectories were reported, and a different specification would have produced different groups with different proportions. The individual assignments are probabilistic, so someone in the 'constantly low' 53% may have been borderline. The design is also selected on death - everyone included has already died, at a mean of 89.8 years, so this describes how function behaved in people who survived to great age, not what to expect from a 70-year-old in clinic today.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.